A Visual Knowledge-Based Navigation for a Robotic Wheelchair

This paper describes and evaluates an intelligent wheelchair, adapted for severely disabled persons with mobility impairement. The wheelchair concept is an assistive device that allows the user to select arbitrary local destinations through a tactile screen interface. The device incorporates an automatic navigation system based on a vision system and knowledge base that drives the wheelchair, avoiding obstacles even in unknown and dynamic scenarios, providing the disabled person with a high degree of autonomy, independent from a particular environment, i.e., not restricted to predefined conditions. In this paper a system is proposed to enable an efficient use of the knowledge base for the perception and navigation of the environment. To support this task, a vision system has been developed in parallel with other sensors. The architecture of the vision system is basically modular to enable a flexible transfer of data between the different modules, and makes any further modification possible and easy. The execution of the different tasks is coordinated by the central module called the supervisor which triggers each module at the appropriate time in order to control the behavior of the wheelchair. One of the most important module in the system is the knowledge base module which uses the acquired and predicted data to construct a scene model. This model is the main interpretation of th environment, since it reflects the perception and prediction processes necessary to a robust and secure navigation. The results indicated that this robotic wheelchair effectively provided mobility and autonomy to the target population.